Search

International audience; The relevance of decision making in autonomous systems is intrinsically related to the system capacity to discriminate its perception-action states. This is particularly challenging in unknown and changing complex environments, where providing a complete a priori representation to the system is not possible. To illustrate the problem, let us consider a decentralized control of road traffic, where a control device of the distributed infrastructure locally controls traffic, by learning to construct a precise representation (perception-action states) of the traffic state. In this context, it is challenging to define from prior knowledge a relevant representation of the traffic state that enables an efficient recommendation-based control. Without considering a prior domain-knowledge representation, we propose an approach able to combine a set of existing traditional unsupervised learning methods that collaborate as a population of agents in order to build an efficient representation. Our approach follows a constructivist learning perspective, where each agent produces a possible discretization of the raw sensed data. Thanks to a multi-agent reinforcement learning process, the population is able to collectively build a representation that combines the good capacities of the individual ones.

International audience; Advances in Information and Communication Technologies (ICT) allow the transportation community to foresee dramatic improvements for the incoming years in terms of a more efficient, environmental friendly and safe traffic management. In that context, new ITS paradigms like Cooperative Systems (C-ITS) enable an efficient traffic state estimation and traffic control. C-ITS refers to three levels of cooperation between vehicles and infrastructure: (i) equipped vehicles with Advanced Driver Assistance Systems (ADAS) adjusting their motion to surrounding traffic conditions; (ii) information exchange with the infrastructure; (iii) vehicle-to-vehicle communication. Therefore, C-ITS makes it possible to go a step further in providing real time information and tailored control strategies to specific drivers. As a response to an expected increasing penetration rate of these systems, traffic managers and researchers have to come up with new methodologies that override the classic methods of traffic modeling and control. In this paper, we discuss some potentialities of C-ITS for traffic management with the methodological issues following the expansion of such systems. Cooperative traffic models are introduced into an open-source traffic simulator. The resulting simulation framework is robust and able to assess potential benefits of cooperative traffic control strategies in different traffic configurations.

In a near future, connected and automated vehicles will progressively replace current vehicles, leading to deep changes in transportation. The driver will be soon assisted and then replaced by an embedded system, able to act quicker, relying on a more robust and precise representation of its surrounding environment. However, some steps are still needed before coming up with such a level of automation since the vehicle environment is complex and unpredictable.This work intends to anticipate the introduction of these new kinds of vehicles by providing cooperative behaviors at both infrastructure and vehicle levels, at the same time allowing a decentralized control of these systems. We propose a distributed modeling framework, using multi-agent systems, relying on the coupling of the system dynamics\string: information, communication and reliability (modeled through the concept of trust).The next step was to develop a simulation framework enabling the implementation of our models for connected vehicles applications. We present a new microscopic traffic simulator, built as an extension of an existing platform, and able to model information exchanges using messages between vehicles and with the infrastructure. All data are provided by sensors and all entities, modeled as agents, are autonomous regarding their decision process. Thanks to the simulator, it is possible to imagine new control strategies relying on recommendations disseminated by the connected infrastructure. Consistency and interdependence of the simulator components are ensured by the dynamic coupling. As for the vehicles' dynamics, we propose a bilateral multi-anticipative model that integrates additional information from communications in the vehicle decision process. Results in simulation confirm that the model is able to reduce the propagation of perturbation through the flow, leading to a more homogeneous and stable traffic.One of the major issues regarding traffic control strategies will be to dynamically adapt the action policy to the several deployment stages of cooperative transportation systems. The similarities with Artificial Intelligence problems like cognition motivate a more abstract study\string: how to model an autonomous system able to control its environment. We choose the constructivist approaches, that propose to model the cognition process as an iterative building process. For cooperative traffic, the benefits lie in the ability of the system to generate its own strategies, relying or not on domain specific knowledge, and then make them evolve to be adapted to vehicles in the flow. The results from our approach are presented in two distinct simulation frameworks. The first one is an experimentation prototype aiming at highlighting the low-level behaviors in a simplified environment. In this context, we show that the model is able to combine efficiently several individual concurrent representations in order to build a high-level representation that can be adapted to several contexts. The second framework is the traffic simulator where the results lead to some insights about the potential of our approach for such realistic applications., Dans un proche futur, les véhicules connectés et autonomes remplaceront nos véhicules actuels, et il sera nécessaire de repenser intégralement la mobilité. Le conducteur, avec ses lacunes, sera de plus en plus assisté, et un jour détrôné par un système embarqué, capable d'agir plus rapidement, tout en ayant une représentation plus précise et fiable de son environnement de navigation. Pourtant, il reste encore du chemin à parcourir avant d'arriver à un tel stade de maturité : l'environnement du véhicule est complexe, imprévisible et conflictuel, car partagé avec d'autres acteurs de la mobilité.Ce travail de thèse vise à anticiper l'arrivée de ces nouveaux véhicules afin de proposer des comportements coopératifs au niveau des véhicules et de l'infrastructure tout en permettant un contrôle décentralisé de ce type de système complexe. Dans le cadre d'une approche multi-agents et d'une architecture distribuée, nous proposons d'abord une modélisation par couplage des dynamiques physique et communicationnelle, auxquelles s'ajoute une intégration de la fiabilité de l'information (confiance).L'étape suivante a été de développer un cadre de simulation propice à l'implémentation de nos modèles dans le cas des véhicules connectés. Nous introduisons un nouveau simulateur de trafic, construit comme une extension d'une plateforme existante, au sein duquel les flux d'informations entre les véhicules et avec l'infrastructure prennent la forme d'échanges de messages. Toutes les informations du système proviennent de capteurs, et toutes les entités, modélisées comme des agents, sont autonomes dans leur prise de décision. De nouvelles formes de contrôle sont désormais envisageablesen utilisant des consignes transmises par l'infrastructure communicante. Le couplage des dynamiques assure la cohérence et l'inter-dépendance des différents modèles dans le simulateur. Nous montrons en simulation que, grâce à l'intégration d'informations supplémentaires via la communication, les véhicules, modélisés par un modèle microscopique multi-anticipatif bilatéral, sont capables de réduire l'effet de perturbations propagées au sein d'un flux. En termes de stratégies de contrôle, une des problématiques principales est de garantir une forme de contrôle qui s'adaptera aux différentes phases de déploiement des systèmes coopératifs. L'analogie avec des problèmes de l'IA (problème de cognition) nous a mené à traiter le problème de manière plus abstraite : comment permettre à un système autonome de contrôler son environnement. Les approches constructivistes, que nous avons retenues, modélisent le processus de cognition comme un phénomène de construction itératif. Pour le trafic coopératif, l'avantage est de disposer d'un système capable de générer ses propres stratégies, en utilisant ou non des connaissances expertes, et de les faire évoluer au cours du temps pour s'adapter aux véhicules composant le flux. Les résultats de notre approche sont présentés dans deux cadres de simulation. Le premier est un prototype visant à illustrer les comportements de bas niveau dans un environnement simplifié. Nous montrons que le modèle est capable dans ce cadre de combiner différentes représentations individuelles pour construire une représentation et de s'adapter à différents contextes en les recombinant dynamiquement. Puis, dans le cadre de simulation du trafic coopératif, les résultats laissent entrevoir le potentiel de notre approche dans des applications réelles.

Spatiotemporal data, and more specifically origin-destination matrices, are critical inputs to mobility studies for transportation planning and urban management purposes. Traditionally, high-cost and hard-to-update household travel surveys are used to produce large-scale origin-destination flow information of individuals' whereabouts. In this paper, we propose a methodology to estimate origin-destination (O-D) matrices based on passively-collected cellular network signalling data of millions of anonymous mobile phone users in the Rhone-Alpes region, France. Unlike Call Detail Record (CDR) data which rely only on phone usage, signalling data include all network-based records providing higher spatiotemporal granularity. The explored dataset, which consists of time-stamped traces from 2G and 3G cellular networks with users' unique identifier and cell tower locations, is used to first analyse the cell phone activity degree indicators of each user in order to qualify the mobility information involved in these records. These indicators serve as filtering criteria to identify users whose device transactions are sufficiently distributed over the analysed period to allow studying their mobility. Trips are then extracted from the spatiotemporal traces of users for whom the home location could be detected. Trips have been derived based on a minimum stationary time assumption that enables to determine activity (stop) zones for each user. As a large, but still partial, fraction of the population is observed, scaling is required to obtain an O-D matrix for the full population. We propose a method to perform this scaling and we show that signalling data-based O-D matrix carries similar estimations as those that can be obtained via travel surveys.

Background Surging acceptance of adaptive cruise control (ACC) across the globe is further escalating concerns over its energy impact. Two questions have directed much of this project: how to distinguish ACC driving behaviour from that of the human driver and how to identify the ACC energy impact. As opposed to simulations or test-track experiments as described in previous studies, this work is unique because it was performed in real-world car-following scenarios with a variety of vehicle specifications, propulsion systems, drivers, and road and traffic conditions. Methods Tractive energy consumption serves as the energy impact indicator, ruling out the effect of the propulsion system. To further isolate the driving behaviour as the only possible contributor to tractive energy differences, two techniques are offered to normalize heterogeneous vehicle specifications and road and traffic conditions. Finally, ACC driving behaviour is compared with that of the human driver from transient and statistical perspectives. Its impact on tractive energy consumption is then evaluated from individual and platoon perspectives. Results Our data suggest that unlike human drivers, ACC followers lead to string instability. Their inability to absorb the speed overshoots may partly be explained by their high responsiveness from a control theory perspective. Statistical results might imply the followers in the automated or mixed traffic flow generally perform worse in reproducing the driving style of the preceding vehicle. On the individual level, ACC followers have tractive energy consumption 2.7 – 20.5 % higher than those of human counterparts. On the platoon level, the tractive energy values of ACC followers tend to consecutively increase (11.2 – 17.3 %). Conclusions In general, therefore, ACC impacts negatively on tractive energy efficiency. This research provides a feasible path for evaluating the energy impact of ACC in real-world applications. Moreover, the findings have significant implications for ACC safety design when handling the stability-responsiveness trade-off.

In the last decade, humanity has faced many different pandemics such as SARS, H1N1, and presently novel coronavirus (COVID-19). On one side, scientists have developed vaccinations, and on the other side, there is a need to propose models that can help in understanding the spread of these pandemics as it can help governmental and other concerned agencies to be well prepared, especially for pandemics, which spreads faster like COVID-19. The main reason for some epidemic turning into pandemics is the connectivity among different regions of the world, which makes it easier to affect a wider geographical area, often worldwide. Also, the population distribution and social coherence in the different regions of the world are non-uniform. Thus, once the epidemic enters a region, then the local population distribution plays an important role. Inspired by these ideas, we propose two versions of our mobility-based SIR model, (i) fully-mixed and (ii) for complex networks, which especially takes into account real life interactions. To the best of our knowledge, this model is the first of its kind, which takes into account the population distribution, connectivity of different geographic locations across the globe, and individuals' network connectivity information. In addition to presenting the mathematical proof of our models, we have performed extensive simulations using synthetic data to demonstrate the generalization capability of our models. Finally, to demonstrate the wider scope of our model, we applied our model to forecast the COVID-19 cases at county level (Estonia) and regional level (Rhône-Alpes region in France).

Call detail records (CDR) collected by mobile phone network providers have been largely used to model and analyze human-centric mobility. Despite their potential, they are limited in terms of both spatial and temporal accuracy thus being unable to capture detailed human mobility information. Network Signaling Data (NSD) represent a much richer source of spatio-temporal information currently collected by network providers, but mostly unexploited for fine-grained reconstruction of human-centric trajectories. In this paper, we present TRANSIT, TRAjectory inference from Network SIgnaling daTa, a novel framework capable of processing NSD to accurately distinguish mobility phases from stationary activities for individual mobile devices, and reconstruct, at scale, fine-grained human mobility trajectories, by exploiting, with a DBSCAN-based clustering approach, the inherent recurrence of human mobility and the higher sampling rate of NSD. The validation on a ground-truth dataset of GPS trajectories showcases the superior performance of TRANSIT (80% precision and 96% recall) with respect to state-of-the-art solutions in the identification of movement periods, as well as an average 190 m spatial accuracy in the estimation of the trajectories. We also leverage TRANSIT to process a unique large-scale NSD dataset of more than 10 millions of individuals and perform an exploratory analysis of city-wide transport mode shares, recurrent commuting paths, urban attractivity and analysis of mobility flows. TRUE pub

Taxis are an important transportation mode in many cities due to their convenience and accessibility. In the taxi-dispatching problem, sometimes it is more beneficial for the supplier if taxis cruise in the network after serving the first request to pick up the next passenger, while sometimes it is better that they wait in stations for new trip requests. In this article, we propose a rolling horizon scheme that dynamically optimizes taxi dispatching considering the actual traffic conditions. To optimize passenger satisfaction, we define a limitation for passenger waiting time. To be able to apply the method to large-scale networks, we introduce a clustering-based technique that can significantly improve the computation time without harming the solution quality. Finally, we test our method on a real test case considering taxi requests with personal car trips to reproduce actual network loading and unloading congestion during peak hours.

Railway operations in dense traffic areas are very sensitive even to small disturbances, and thus require careful planning and real-time management. Dwell times in stations are in particular subject to a high variability and are hard to predict; this is mostly due to the interactions between passengers and the railway system during the dwelling process. This paper presents a data-driven approach for assessing the influence of the numbers of alighting, boarding and on board passengers on the dwell time. We propose to split the dwell time into a deterministic component depending on the passenger flow, called the Minimum Dwell Time, and a random component. A method for estimating the minimum dwell time is provided. Based on the knowledge of this value, observations can be classified according to the main determinant of dwell time, namely timetable constraints or passenger exchange. The latter observations are used for estimating the conditional distribution of dwell time given passenger flows. Numerical experiments are carried out on stations located inside the dense traffic area of Paris suburban network. The obtained results indicate that the presented method can be used for a variety of applications, such as capacity assessment or stochastic simulation.

Mobile phone data collected by network operators can provide fundamental insights into individual and aggregate mobility of people, at unprecedented spatiotemporal scales. However, traditional call detail records (CDR) have fundamental issues because of low accuracy in both spatial and temporal dimensions, which limits their applicability for detailed studies on mobility, especially in urban scenarios. This paper focuses on a new generation of mobile phone passive data, individual cellular network signaling data, characterized by higher spatiotemporal resolutions than traditional CDR. A framework based on unsupervised hidden Markov model is designed for map-matching such data on a multimodal transportation network, aimed at accurately inferring the complex multimodal travel itineraries and popular paths people follow in their urban daily mobility. This information, especially if computed at large spatiotemporal scales, can represent a solid basis for studying actual and dynamic travel demand, to properly dimension multimodal transport systems and even perform anomaly detection and adaptive network control. The approach is evaluated in a case study based on real cellular traces collected by a major French operator in the city of Lyon, and a validation study at both microscopic and macroscopic levels proposed. The results show that this approach can properly handle sparse and noisy cell phone trajectories in complex urban environments. Moreover, the results are promising concerning popular paths detection and reconstruction of origin–destination matrices.

Datasets of mobile phone trajectories collected by network operators offer an unprecedented opportunity to discover new knowledge from the activity of large populations of millions. However, publishing such trajectories also raises significant privacy concerns, as they contain personal data in the form of individual movement patterns. Privacy risks induce network operators to enforce restrictive confidential agreements in the rare occasions when they grant access to collected trajectories, whereas a less involved circulation of these data would fuel research and enable reproducibility in many disciplines. In this work, we contribute a building block toward the design of privacy-preserving datasets of mobile phone trajectories that are truthful at the record level. We present GLOVE, an algorithm that implements k-anonymity, hence solving the crucial unicity problem that affects this type of data while ensuring that the anonymized trajectories correspond to real-life users. GLOVE builds on original insights about the root causes behind the undesirable unicity of mobile phone trajectories, and leverages generalization and suppression to remove them. Proof-of-concept validations with large-scale real-world datasets demonstrate that the approach adopted by GLOVE allows preserving a substantial level of accuracy in the data, higher than that granted by previous methodologies. This work was supported by the Atracción de Talento Investigador program of the Comunidad de Madrid under Grant No. 2019-T1/TIC-16037 NetSense.

International audience; A key question about cooperative vehicle longitudinal control is reactivity, which determines the future of road safety, and capacity. In adaptive cruise control (ACC), the controller adapts the speed of the vehicle to its immediate leader's speed whereas, in the cooperative version (CACC), connectivity between the platoon equipped vehicles reduces their response times. The USDoT Cooperative Automated Research Mobility Applications (CARMA) platform provides data for platooning experiments involving ACC and CACC vehicles. We measure ACC response times (mean = 2.78 seconds) larger than for human-driven cars. We study response times inside CACC platoons showing that connectivity is not always effective.

Betweenness Centrality (BC) has proven to be a fundamental metric in many domains to identify the components (nodes) of a system modelled as a graph that are mostly traversed by information flows thus being critical to the proper functioning of the system itself. In the transportation domain, the metric has been mainly adopted to discover topological bottlenecks of the physical infrastructure composed of roads or railways. The adoption of this metric to study the evolution of transportation networks that take into account also the dynamic conditions of traffic is in its infancy mainly due to the high computation time needed to compute BC in large dynamic graphs. This paper explores the adoption of dynamic BC,i.e.,BC computed on dynamic large-scale graphs, modeling road networks and the related vehicular traffic, and proposes the adoption of a fast algorithm for ahead monitoring of transportation networks by computing approximated BC values under time constraints. The experimental analysis proves that, with a bounded and tolerable approximation, the algorithm computes BC on very large dynamically weighted graphs in a significantly shorter time if compared with exact computation. Moreover, since the proposed algorithm can be tuned for an ideal trade-off between performance and accuracy, our solution paves the way to quasi real-time monitoring of highly dynamic networks providing anticipated information about possible congested or vulnerable areas. Such knowledge can be exploited by travel assistance services or intelligent traffic control systems to perform informed re-routing and therefore enhance network resilience in smart cities.

hEART 2020, 9th Symposium of the European Association for Research in Transportation - Virtual conference, Lyon, FRANCE, 03-/02/2021 - 04/02/2021; This work focuses on dynamic sustainable system optimum (SSO). When carrying out dynamic SSO assignment, instead of minimizing their own travel cost of each user, we aim to define an SSO status to minimize the on-road traffic emissions of the whole system. The method is based on an system-optimization program. This work proposes a simple, intuitive, simulation-based methodological framework for solving the SSO problem. The main contribution of this paper is to propose a statistically reliable learning process of link marginals that permits to derive link performance function from any DTA simulator. Then the obtained functions can be used to compute network equilibrium including SSO. A test case is carried out on a toy network. Numerical results show that with the help of the proposed method, we can reduce 5.79% of carbon monoxide (CO) emission on the whole system under SSO equilibrium, when compared with UE.

hEART 2020, 9th Symposium of the European Association for Research in Transportation - Virtual conference, Lyon, France, 03-/02/2021 - 04/02/2021; The present work proposes a framework to calibrate MFD models using mobile phone data. The three major components to calibrate in the present context include MFD shape, regional trip lengths and path flow distribution. Time dependent penetration rates are estimated by fusing the OD matrix and the Loop Detector Data (LDD). The estimated MFDs are stable with very low scatter. In the following step, macro-paths and their corresponding trip lengths are estimated. Dynamic evolution of trip lengths is demonstrated using the present data, which is otherwise very difficult to capture with other types of data sources. Another important component calibrated is the time dependent path flow distribution between the different macropaths for a given OD pair. The manuscript is concluded by presenting the time evolution of the User Equilibrium (UE) gap for different macroscopic OD pairs.

TRB 2021, Transportation Research Board 100th Annual Meeting - A Virtual Event, Washington DC, ETATS-UNIS, 05-/01/2021 - 29/01/2021; The design of network-wide schemes requires computationally efficient traffic models. The Macroscopic Fundamental Diagram (MFD) is a promising tool for such an application. Unfortunately, current semi-analytical approaches require an inaccurate network reduction to a corridor. Our methodology allows us to estimate the MFD for general networks, without the information loss induced by the reduction of networks to corridors. The model is based on the method of cuts, but can account for different demand patterns, and determine the upper bound of network flow. Thereby, we consider flow conservation and the effects of spillbacks at the network level. Furthermore, we propose an overall framework which decomposes the network into a set of corridors, and then applies our model to each of them while taking into account the dependencies across corridors (e.g. due to turning flows and spillbacks). Aggregating the results leads then to a network MFD. The proposed framework applies to any general network, and we show a proof of concept using a simple but sufficient case study. We compare the results to the original method of cuts, and a ground truth derived from a microscopic simulation. Analysis reveals a strongly reduced error of the estimated MFD based on our method, compared to the currently applied approach. The potential of this methodology lies in its implementation simplicity and reduced computational cost.

TRB 2021, Transportation Research Board 100th Annual Meeting - A Virtual Event, Washington DC, ETATS-UNIS, 05-/01/2021 - 29/01/2021; Departure time choice models have a crucial role in transportation network planning in order to provide planners with predictions about the optimal traffic load. This paper introduces a new dynamic departure time choice framework based on the so-called Mean-field Games (MFGs) theory proposed by Lasry and Lions. MFGs models aim to analyze large population differential games. This feature gives us a unique opportunity to design a departure time choice equilibrium model in a scale-free transportation system. The proposed framework is the combination of two components: (i) the reaction of travelers to the traffic congestion by choosing their departure time to optimize the travel cost; (ii) the aggregation of the actions of the travelers, which determines the congestion level of the system. The first component corresponds to a classical game theory model, while the second one captures the interactions of the travelers at the macroscopic level. We use the generalized bathtub model recently proposed by Jin to represent the traffic dynamics at the network level. First, we provide a continuous model to investigate the equilibria and then give a discrete approximation of the system from deterministic differential games models. Finally, we show on a real test case that the model can represent the departure time choice equilibrium for multiple preferred travel time with heterogeneous trip lengths wile the framework is capable of considering a large number of players.

This paper investigates the accuracy of mean density estimation from direct sensing at link and network levels. Different calculation methods are compared depending on sensor type, probe vehicles or loop detectors, and availability to quantify the magnitude of expected errors. Probe data are essential to reduce the error but accurate density estimation requires high penetration rates, which is hardly true in practice. We enhance the fishing rate method, i.e. using the ratio of probes detected at the loop locations over the loop flow, to estimate density. Accurate density estimation at the link level can only be obtained when probes and loop data are available in real-time. At the network level, accurate density estimations can be obtained when combining loop and probe observations, even if few links capture both data sources. It requires applying the proper analytical formulation to aggregate the local observations, i.e. carefully defining fishing rates at this scale.

Urban road transportation performance is the result of a complex interplay between the network supply and the travel demand. Fortunately, the framework around the macroscopic fundamental diagram (MFD) provides an efficient description of network-wide traffic performance. In this paper, we show how temporal patterns of vehicle traffic define the performance of urban road networks. We present two high-resolution traffic datasets covering a year each. We introduce a methodology to quantify the similarity of macroscopic traffic patterns. We do so by using the concepts of the MFD and a dynamic time warping (DTW) based algorithm for time series. This allows us to derive a few representative MFD clusters that capture the essential macroscopic traffic patterns. We then provide an in-depth analysis of traffic heterogeneity in the network which is indicative of the previously found clusters. Thereupon, we define a parsimonious classification approach to predict the expected MFD clusters early in the morning with high accuracy.

COMPLEX NETWORKS 2020, The Ninth International Conference on Complex Networks and their Applications, Madrid, ESPAGNE, 01-/12/2020 - 03/12/2020; With the recent and continuous growth of large metropolis, the development, management and improvement of their urban multi-modal transport networks become a compelling need. Although the creation of a new transport mode often appears as a solution, it is usually impossible to construct at once a full networked public transport. Therefore, there is a need for efficient solutions aimed at prioritizing the order of construction of the multiple lines or transport modes. Hence, the proposed work aims at developing a simple and quick-to-compute methodology aimed at prioritizing the order of construction of the lines of a newly designed transport mode by maximizing the network performance gain, as described by complex networks metrics. In a resilience context, the proposed methodology could also be helpful to support the rapid and quick response to disruptions by setting up or reinforcing an adapted emergency transport line (e.g., bus service) over a set of predefined itineraries.

hEART 2020, 9th Symposium of the European Association for Research in Transportation - Virtual conference, Lyon, FRANCE, 03-/02/2021 - 04/02/2021; This paper investigates the accuracy of accumulation estimation at different urban scales, mainly link and network levels. Different methods are compared depending on sensor types, probe vehicles or loop detectors, and historical data availability. Probe vehicles can provide an accurate estimation of the average travel times but only for a given vehicle sample. So, we resort to loops to calculate the fishing rate and scale up the observations to all vehicles. The fishing rate is the ratio of probes that are detected at the loop locations over the loop flow. Accurate accumulate estimation at the link level can only be obtained when probes and loop data are available in real-time. Methods based on spatially aggregated fishing rates when probe data are available but not loops show better results, but the errors are still significant at the link level. It is possible to obtain accurate accumulation estimations at the network level when combining loop and probe observations, even if few links capture both data sources. It requires applying proper aggregation technics and a careful definition of fishing rates at this scale.

This paper proposes a framework to extract dynamic trip flows and travel demand patterns from large-scale 2 G and 3 G cellular signaling data. Novel data pre-processing techniques based on cell phone activity metrics are presented. The trip extraction method relies on the detection of stationary activities to form trip sequences related to resident users. A probabilistic solution is introduced to estimate the trip starting time, allowing to aggregate trips by time of the day and reconstruct hourly travel flows. To better characterize these flows, a spatial clustering process combined with land-use data is proposed based on the temporal demand profile of each zone. Empirical comparisons have been performed showing that the resulting dynamic travel demand patterns are consistent with those obtained from travel survey data with high correlation coefficients of about 0.9. The results prove the potential of signaling data to generate low-cost valuable information for large-scale travel demand modeling.

Since the introduction of the Macroscopic Fundamental Diagram (MFD), many traffic control strategies and algorithms have been developed to implement MFD-based perimeter control over a specific urban region. A model-based controller consists of two components: a plant model that represents reality; and a prediction model used to determine optimal control actions. In most studies, the authors assume a constant average trip length for all drivers traveling within the same region, for the prediction model. In these studies about perimeter control and MFD traffic models, the controllers show a good performance because accumulations, i.e. traffic states, from the plant are used to reflect the initial state of the prediction model with a high frequency (about a few seconds). However, this average trip length changes over time as it depends on the Origin-Destination flow decomposition, playing an important role in real applications. The main contributions of this paper are twofold. First, we show that the assumption about constant trip lengths used in the prediction model deteriorates the controller's performance for low frequency updates of the optimal control actions. Second, we propose a methodological framework based on the Unscented Kalman Filter (UKF) for dynamically adjusting the average trip lengths and accumulations. Our test results on a real city network show that applying this methodological framework significantly improves the controller's performance.

International audience; This paper presents a new concept for operating traffic management at a sizeable urban scale. The overall principle is to partition the network into multiple regions, where traffic conditions should remain optimal. Instead of monitoring the inflow, like for perimeter control, deviations in regional mean speed compared to the reference are transformed into avoidance levels by a centralized controller. That information is broadcasted to vehicles through a public avoidance map. The onboard navigation systems then interpret this map to deliver individual route guidance according to the avoidance levels. In that way, the concept tends to distribute vehicles in the network optimally while preserving privacy. In addition to the controller parameters itself, three other critical parameters define the system: the safety distance, the controller time horizon, and the region sizes. Thorough sensitivity analysis leads to the optimal setting. The concept is proven effective using microsimulation for both a Manhattan and a realistic network. The total travel times is improved by about 15% when traffic is severely congested compared to the uncontrolled case. In the meantime, the mean individual travel distance increase for rerouted vehicles is kept below 10%. Those results have been obtained with a simple decentralized reactive control framework, i.e., an individual proportional feedback system independently governs each region. More advanced formulations introducing cooperation between the regions are also tested. In a nutshell, this paper is a proof of concept for a new control system that appears both practical and valuable to alleviate congestion in urban areas

Transport networks are essential for societies. Their proper operation has to be preserved to face any perturbation or disruption. It is therefore of paramount importance that the modeling and quantification of the resilience of such networks are addressed to ensure an acceptable level of service even in the presence of disruptions. The paper aims at characterizing network resilience through weighted degree centrality. To do so, a real dataset issued from probe vehicle data is used to weight the graph by the traffic load. In particular, a set of disrupted situations retrieved from the study dataset is analyzed to quantify the impact on network operations. Results demonstrate the ability of the proposed metrics to capture traffic dynamics as well as their utility for quantifying the resilience of the network. The proposed methodology combines different metrics from the complex networks theory (i.e., heterogeneity, density, and symmetry) computed on temporal and weighted graphs. Time-varying traffic conditions and disruptions are analyzed by providing relevant insights on the network states via three-dimensional maps.

With the recent and continuous growth of large metropolis, the development, management and improvement of their urban multi-modal transport networks become a compelling need. Although the creation of a new transport mode often appears as a solution, it is usually impossible to construct at once a full networked public transport. Therefore, there is a need for efficient solutions aimed at prioritizing the order of construction of the multiple lines or modes that a transport operator might want to construct to increase its offer. For this purpose, we propose in this paper a simple and quick-to-compute methodology, called REINFORCE, to prioritize the order of construction of the lines of a newly designed transport mode by maximizing the transport network performances and enhancing the transport network resilience, as described by complex networks metrics. REINFORCE could also be helpful to support the rapid and quick response to disruptions by setting up or reinforcing an adapted emergency transport line (e.g., bus service) over a set of predefined itineraries.

Nowadays a large amount of data is originated by complex systems, such as social networks, transportation systems, computer and service networks. These systems can be modeled by using graphs and studied by exploiting graph metrics, such as Betweenness Centrality (BC), a popular metric to analyze node centrality of graphs. In spite of its great potential, this metric requires long computation time, especially for large graphs. In this paper, we present a very fast algorithm to compute BC of undirected graphs by exploiting clustering. The algorithm leverages structural properties of graphs to find classes of equivalent nodes: by selecting one representative node for each class, we are able to compute BC by significantly reducing the number of single-source shortest path explorations adopted by Brandes' algorithm. We formally prove the graph properties that we exploit to define the algorithm and present an implementation based on Scala for both sequential and parallel map-reduce executions. The experimental evaluation of both versions, conducted with synthetic and real graphs, reveals that our solution largely outperforms Brandes' algorithm and significantly improves known heuristics.

Multiple factors can influence the public transport level of service. All take root in the network structure and the operating regime, i.e. how the bus lines are arranged atop the street network and how the service frequency is adjusted to meet urban mobility patterns. This is known as the bus network design problem and has been the subject of several studies. The problem is so challenging that most studies until now resort to strong assumptions such as a static description of the peak hour demand, homogeneous user behavior, and equal trip lengths. Potential effects of different types of user behavior and trip lengths patterns on the user and/or operator cost have not been investigated whatsoever. Moreover, none of the existing studies have considered the effects of bus network structure on private car users, the level of interactions between the modes, and the passenger mode choice that depends on traffic conditions. This paper aims to close this gap and provide a general framework considering multiple trip length patterns, two types of user behavior, and the effects that the bus network structure might have on the traffic performance and passenger mode choice. For modeling different trip length patterns, we use the trip length distribution as an intermediate level of abstraction. To capture complex modal interactions and quantify the operating speeds, we apply the recently proposed three-dimensional macroscopic fundamental diagram. We use the operating speed for each mode to determine the mode choice at the trip length level. This way, we are able to solve the optimal bus network design problem under the free-flow/saturated traffic conditions in an analytical way, while considering more realistic settings including a dynamic description of the peak hour demand, mixed traffic, and different mode choice decisions depending on trip lengths and walking preferences. Numerical analysis reveals that all the tested factors, including demand intensity, user behavior, and trip length patterns, have significant effects on the operator and user cost function. Results show that the probability of choosing any given mode follows certain distribution that varies across the trip length patterns, indicating the importance of modeling the mode choice at the trip length level. Furthermore, the results indicate that users can benefit if they are willing to adjust the number of transfers to minimize the walking distance at the origin and the destination. Moreover, we show that the optimal bus network design determined for the uniform trip pattern underestimates the number of required buses, which leads to passenger congestion at stops during the peak period. This, however, does not happen when we take into account the actual trip length distribution for the bus network design. A comparison with a simplified approach that considers the bus system only, reveals the value of accounting for the complex bi-modal interactions, especially for higher demand levels. Finally, we show that by allowing the design parameters to vary across cardinal directions we provide more flexibility for the bus system to serve the passenger demand while reducing the operator cost compared to the existing approaches. ISSN:0191-2615 ISSN:1879-2367

TRB 2021, Transportation Research Board 100th Annual Meeting - A Virtual Event, Washington DC, ETATS-UNIS, 05-/01/2021 - 29/01/2021; The work proposes a multi-modal regional mean speed regression analysis for the city network of Athens, Greece. The dataset from pNUEMA experiment is used in the present context. Accumulations and mean speeds of different modes are estimated and compared to each other. Besides, the mean running speeds of all modes are also computed and compared. The mean speed of each mode is assumed to be a linear combination of accumulations of all modes and regression analysis is performed. Ordinary Least Squares (OLS) and Partial Least Squares (PLS) methods are used in the current work. It is noticed that multicollinearity between different modes is weak in the current dataset. The quality of multi-modal regression fits is compared to the uni-modal ones, where the mean speed of a given mode is assumed to be a function of the accumulation of that mode only. It is concluded that multi-modal regression fits outperform their uni-modal counterparts in terms of R2 and Root Mean Squared Relative Error (RMSRE).

Over the past few decades, the digitalization of services and infrastructures has led to the emergence of a broad set of new information sources to characterize human mobility. These sources usually offer valuable significant population penetration rates but may also suffer from important temporal sparsity. Data generated by user activity, such as social networks or mobile phone data, especially fit this description. Although this temporal sparsity might prevent estimating individual travel speeds, we state that such low-frequency positioning data enable estimating the average urban traffic speed dynamics when considering an adequate network partitioning. In this sense, this article proposes a new method, based on the division of the urban area of a given city into regions and on the analysis of a limited set of basic characteristics of individual vehicle trips, such as the regional path. Our solution first involves estimating robust travel times from travelers sharing similar trip features and then jointly analyzing these travel times to deduce the underlying regional traffic speeds, using regression analysis. We apply this methodology on a set of trips derived from a large GPS dataset of vehicle tracks covering the city of Lyon. These data are purposely downsampled to reduce the sampling rate and reproduce bias and temporal features that are proper to sparser but larger-scale, mobility data sources dependent on user’s communication activities. Controlling the data downsampling process allows us to evaluate the impacts of the progressive information loss on the speed estimation, while the raw GPS data provide the ground truth speed reference against which to compare our results. Provided that the amount of observed individual trips is sufficient, the analysis returns satisfying speed estimation results, both at low and high downsampling levels. Thus, we successfully demonstrate that it is possible to estimate zonal traffic speeds from degraded trip data without evaluating individual travel speeds.

This work investigates network-related trajectory features to unravel trips that contribute most to system under-performance. When such trips are identified, feature analysis also permits determining the best alternatives in terms of routes to bring the system to its optimum. First, we define a combination of network-related trajectory features that helps us unravel the critical trips which contribute the most to the network under-performance, based on the literature review and a factor selection process. Second, based on supervised learning methods, we propose a two-step data-driven methodological framework to reroute a part of the users and make the system close to its optimum. The learning models are trained with trajectory features to identify which users should be selected, and which alternative routes should be assigned, thanks to the data and features obtained from two reference dynamic traffic assignment (DTA) simulations, under User-Equilibrium (UE) and System-Optimum (SO). We only focus on trajectory features that are accessible in real time, such as network features and regular travel time information, so that the methods proposed can be implemented without requiring cumbersome network monitoring and prediction. Finally, we evaluate the efficiency of the methods proposed through microscopic DTA simulations. The results show that by targeting 20% of the users according to our selection model and moving them onto paths predicted as optimal alternative paths based on our rerouting model, the total travel time (TTT) of the system is reduced by 5.9% in comparison to a UE DTA simulation. This represents 62.5% of the potential TTT reduction from UE to SO, when all the users choose their path under the SO condition.

International audience

One of the main challenges for multi-regional application of the aggregated traffic models based on the Macroscopic Fundamental Diagram, lies in the identification and characterization of the most prevailing paths chosen by drivers. In this paper, we propose a methodological framework, based on two distinct methods, to determine these prevailing paths. The first method requires the information about travel patterns in the urban network as well as the information about the city network partitioning. The second method is more parsimonious, and consists on the direct calculation of shortest-cost paths on the aggregated network. For this, we propose four impedance functions that utilize topological features of the urban network and its partitioning. We test the performance of this methodological framework for determining the most prevailing paths on a network representing the metropolitan area of Lyon (France). We consider a set of real trajectories (i.e. GPS data) of drivers in this network as a benchmark. We show that the proposed methods are able to identify the most prevailing paths as the ones chosen by drivers, as evidenced by a large similarity value between the sets of paths. Based on a maximum likelihood estimation, we also show that the Weibull distribution is the one that better reproduces the functional form of the network-wide distribution of travel distances. However, the characterization of the functional form of such distributions characteristic to each region defining a path is not trivial, and depends on the complex topological features of the urban network concerning the definition of its partitioning. We also show that the Euclidean distance metrics provides good estimates of the average travel distances. Interestingly, we also show that the most prevailing paths are not necessarily the ones that have the lowest average travel distances.

Recent empirical studies have found that travellers route choices deviate from perfect rationality, showing that urban trips do not necessary follow the shortest-time routes. However, there is no consent on how much the travellers' route choice behaviour deviates from the perfect rational assumption. The objective of this study is to contribute to the understanding on how travellers process travel time when making route choices, and to quantify to what extent users are strict travel time minimisers or if bounded rationality is observed. The question of whether travellers evaluate travel time differences in absolute or relative terms is also addressed, and the heterogeneity in the route choice behaviour of travellers investigated. The results of route choice experiments, focused on the choices in diverse OD pairs and traffic conditions, are analysed. In total, 496 participants recorded 5535 choices over 41 OD pairs. It was found that travellers evaluate relative rather than absolute differences in travel time. In 60.5% of the trips participants chose the fastest route, but this percentage is 80% when the travel time between the fastest and the rest of the alternatives is at least 30% higher. Only 10% of the individuals chose the fastest route in all trips, confirming the hypothesis of bounded rationality. Participants exhibited heterogeneous travel time indifference bands: the average participant was indifferent to relative travel time differences of less than 31%. A mixed logit model (MXL), considering heterogeneous indifference bands is proposed. The model shows a similar predictive accuracy compared to the classical MXL model.

This study reviews existing computational methods to calculate simulation-based dynamic network equilibrium. We consider a trip-based multi-modal approach for the dynamic network loading. Mode and path choices are carried out at the same level; therefore, travel times depend on the travel path and the mode attributes of travelers. This study develops a multi-class model with several parameters per class. Two different categories of algorithms (heuristic and meta-heuristic) are considered in order to solve the discrete DTA problem. Finally, we analyze the equilibrium in a large-scale multi-modal DTA test case (Lyon 6th + Villeurbanne) in order to investigate the performance of different optimization approaches to solve trip-based DTA. The results show that, in a multimodal and heterogeneous setting, the meta-heuristic methods provide better solutions than the heuristic methods in terms of optimality and computation time. These improvements are even more significant than in a homogeneous setting.

Can dynamic ride-sharing reduce traffic congestion? In this paper we show that the answer is yes if the trip density is high, which is usually the case in large-scale networks but not in medium-scale networks where opportunities for sharing in time and space become rather limited. When the demand density is high, the dynamic ride-sharing system can significantly improve traffic conditions, especially during peak hours. Sharing can compensate extra travel distances related to operating a mobility service. The situation is entirely different in small and medium-scale cities when trip shareability is small, even if the ride-sharing system is fully optimized based on the perfect demand prediction in the near future. The reason is simple, mobility services significantly increase the total travel distance, and sharing is simply a means of combating this trend without eliminating it when the trip density is not high enough. This paper proposes a complete framework to represent the functioning of the ride-sharing system and multiple steps to tackle the curse of dimensionality when solving the problem. We address the problem for two city scales in order to compare different trip densities. A city scale of 25 km 2 with a total market of 11,235 shareable trips for the medium-scale network and a city scale of 80 km2 with 205,308 demand for service vehicles for the large-scale network over a 4-hour period with a rolling horizon of 20 minutes. The solutions are assessed using a dynamic trip-based macroscopic simulation to account for the congestion effect and dynamic travel times that may influence the optimal solution obtained with predicted travel times. This outperforms most previous studies on optimal fleet management that usually consider constant and fully deterministic travel time functions.

TRB 2021, Transportation Research Board 100th Annual Meeting - A Virtual Event, Washington DC, ETATS-UNIS, 05-/01/2021 - 29/01/2021; An excessive demand of vehicles to a motorway bottleneck leads to traffic jams. Motorbikes are narrow and can drive next to each other in a lane, or in-between lanes in low speeds. This paper analyses the resulting traffic characteristics and presents numerical scheme for a macroscopic traffic flow model for these two classes. The behavior included is as follows. If there are two motorbikes behind each other, they can travel next to each other in one lane, occupying the space of one car. Also, at low speeds of car traffic, they can go in between the main lanes, creating a so-called filtering lane. The paper numerically derives functions of class-specific speeds as function of the density of both classes, incorporating flexible lane usage dependent on the speed. The roadway capacity as function of the motorbike fraction is derived, which interesting can be in different types of phases (with motorbikes at higher speeds or not). We also present a numerical scheme to analyse the dynamics of this multi-class system. We apply the model to an example case, revealing the properties of the traffic stream , queue dynamics and class specific travel times. The model can help in showing the relative advantage in travel time of switching to a motorbike.

Analysing the properties of a network equilibrium can help to have a better view about network state, robustness, and the effect of any variation in the network. This study investigated the impacts of network design history on day-to-day multimodal user equilibrium. In particular, we analyze the long-term evolution of the network, including opening new multimodal options and its impacts on the final network equilibrium. First, the analysis focuses on static network loading with different successive configurations. Then, a more realistic setting is studied by simulation. A large-scale multimodal network with the flexible opening over time of three possible transport facilities shows that the final equilibrium is not unique; more importantly, significant differences can be observed in public transportation occupancy, while user equilibrium is enforced in all situations. Some solutions prove to be better from the collective viewpoint (shorter total travel time), thus giving new insight into public transport planning.

The estimation of trip lengths has been proven to be a key feature for the application of aggregated traffic models based on the Macroscopic Fundamental Diagram. The paths and distances to be traveled by vehicles in regional networks vary over time, due to changes in the traffic conditions. In this paper, we develop a methodological framework to explicitly determine traffic-dependent regional paths and estimate their travel distances. This framework is incorporated into a dynamic traffic assignment module designed to target the Deterministic and Stochastic User Equilibrium in regional networks. We first discuss how regional paths and their characteristic trip lengths are influenced by changes in the regional traffic dynamics. We then test the proposed methodology for estimating traffic-dependent travel distances on small and medium-sized networks, considering a simulation environment. We show that our methodology provides good estimations of the traffic-dependent trip lengths. Our results also shed light on the importance of how time-dependent trip lengths influence the traffic dynamics in the regions.

International audience; Simulation-based traffic impact assessment studies of advanced technologies such as truck platooning need to be carried out to ascertain their benefits for traffic efficiency, safety and environment. To reduce uncertainty in the results of such simulation-based studies, the same simulation studies can be performed in different simulation software. Many traffic simulation software packages (Aimsun, SymuVia, Vissim, SUMO) are currently available for traffic impact assessment of new technologies such as truckplatooning. However, to fully model and simulate the functionalities of such advanced technologies in different simulation environments, several extensions need to be made to the simulation platforms. In most cases, these extensions have to be programmed in different programming languages (C++, Python) and each simulator has its own simulator specific API. This makes it difficult to reuse software written for a specific functionality in one simulation platform in a different simulation platform. To overcome this issue, this paper presents a novel architecture for cross-platform simulation. The architecture is designed such that a specific functionality such as truck-platooning or any other functionality is made platform independent. We designed a cross-platform architecture for simulating a truck-platooning functionality using Vissim and SymuVia simulation software to determine the traffic flow effects of multi-brand truck platooning in the context of the EU project ENSEMBLE. In this draft paper, we present the structure of the framework as well as some preliminary results from a simple simulation performed with the cross-platform simulator.

The discrete network design problem (DNDP) is a well-studied bilevel optimization problem in transportation. The goal of the DNDP is to identify the optimal set of candidate links (or projects) to be added to the network while accounting for users’ reaction as governed by a traffic equilibrium. Several approaches have been proposed to solve the DNDP exactly using single-level, mixed-integer programming reformulations, linear approximations of link travel time functions, relaxations and decompositions. To date, the largest DNDP instances solved to optimality remain of small scale and existing algorithms are no match to solve realistic problem instances involving large numbers of candidate projects. In this work, we examine the literature on exact methodologies for the DNDP and attempt to categorize the main approaches employed. We introduce a new set of benchmarking instances for the DNDP and implement three solution methods to compare computational performance and outline potential directions for improvement. For reproducibility purposes and to promote further research on this challenging bilevel optimization problem, all implementation codes and instance data are provided in a publicly available repository.

4th webinar Ectri Traffic Management Thematic Group, Bron, FRANCE, 08-/10/2020 - 08/10/2020; We realized an analysis of the CACC data of the Volpe laboratory. We observe that the response times distributions are spreading when the distance to the leader and the position in the platoon increase. This leads us to question the connectivity inside those platoons.

The main goal of this dissertation is to answer one of the critical questions about dynamic ride-sharing services: Can dynamic ride-sharing reduce congestion? In this thesis, we propose a simulation-based optimization framework for dynamic ride-sharing. Then using this framework, we assess the dynamic ride-sharing impact on two different network scales to find the answer to this question. When assessing the dynamic ride-sharing problem, two important points should be considered. First, how the ride-sharing system serves the network demand and second, how the ride-sharing system is impacted by the network and in particular by congestion. Then we can assess the impact of such a service on the network. Most of the existing approaches focus on the first point, i.e., designing the demand matching while using basic assumptions for the second point, mainly constant travel times. The proposed method in this thesis can outperform the existing methods in the literature. The optimization algorithm can provide high-quality solutions in a short time. Our solution approach is designed to be exact for small samples. Then, to be able to handle the large-scale problems, it is extended with several heuristics that keep the general design for the solution method but significantly reduce its computation time. In the simulation component, a "Plant Model" is applied based on the "Trip-based Macroscopic Fundamental Diagram (MFD)" to represent the traffic dynamics reality and a "Prediction Model" is applied based on the mean-speed to be used during the assignment process. We perform an extensive simulation study (based on real-world traffic patterns) to assess the influence of dynamic ride-sharing systems on traffic congestion. In the medium-scale (Lyon 6 + Villeurbanne), the results showed that ride-sharing could not significantly improve the traffic situation. High levels of market-share add additional travel distance and travel time to the trips and lead to more traffic in the network. In large cities, the results are entirely different from those in small and medium-sized cities. In large-scale (Lyon city in France) simulations, the proposed dynamic ride-sharing system can significantly improve traffic conditions, especially during peak hours. Increasing the market-share and the number of sharing can enhance this improvement. Therefore, the proposed dynamic ride-sharing system is a viable option, alleviating stress on existing public transport, to reduce the network traffic in populated and large-scale cities., L'objectif principal de cette thèse est de répondre à l'une des questions essentielles sur les services de covoiturage dynamique : le covoiturage dynamique peut-il réduire la congestion ? Dans cette thèse, nous proposons un framework d'optimisation basé sur la simulation pour le covoiturage dynamique. Ensuite, à l'aide de ce framework, nous évaluons l'impact du covoiturage dynamique sur deux échelles de réseau différentes pour trouver la réponse à cette question. Lors de l'évaluation du problème du covoiturage dynamique, deux points importants doivent être pris en compte. Premièrement, comment le système de covoiturage répond à la demande du réseau et deuxièmement, comment le système de covoiturage est influencé par le réseau et en particulier par la congestion. Ensuite, nous pouvons évaluer l'impact d'un tel service sur le réseau. La plupart des approches existantes se concentrent sur le premier point, c'est-à-dire sur la conception de la correspondance de la demande, tout en utilisant des hypothèses de base pour le second point, principalement des temps de trajet constants. La méthode proposée dans cette thèse peut être plus performante que les méthodes existantes dans la littérature. L'algorithme d'optimisation peut fournir des solutions de haute qualité en peu de temps. Notre approche de solution est conçue pour être exacte pour les petits échantillons. Ensuite, pour pouvoir traiter les problèmes à grande échelle, elle est complétée par plusieurs heuristiques qui conservent la conception générale de la méthode de solution mais réduisent considérablement son temps de calcul. Dans la composante de simulation, un "Modèle d'usine" est appliqué sur la base du "Diagramme macroscopique fondamental (MFD) basé sur les déplacements" pour représenter la réalité de la dynamique du trafic et un "Modèle de prédiction" est appliqué sur la base de la vitesse moyenne à utiliser pendant le processus d'affectation. Nous réalisons une étude de simulation approfondie (basée sur des modèles de trafic réels) pour évaluer l'influence des systèmes de covoiturage dynamiques sur la congestion du trafic. À l'échelle moyenne (Lyon 6 + Villeurbanne), les résultats ont montré que le covoiturage ne pouvait pas améliorer de manière significative la situation du trafic. Des parts de marché élevées augmentent la distance et la durée des trajets et entraînent une augmentation du trafic sur le réseau. Dans les grandes villes, les résultats sont totalement différents de ceux des villes petites et moyennes. Dans les simulations à grande échelle (ville de Lyon en France), le système de covoiturage dynamique proposé peut améliorer considérablement les conditions de circulation, en particulier aux heures de pointe. L'augmentation de la part de marché et du nombre de covoiturages peut renforcer cette amélioration. Par conséquent, le système de covoiturage dynamique proposé est une option viable pour réduire la pression sur les transports publics existants et diminuer le trafic du réseau dans les villes peuplées et de grande taille.

IEEE ITSC 2020, 23rd IEEE International Conference on Intelligent Transportation Systems, Rhodes, GRECE, 20-/09/2020 - 23/09/2020; In order to provide a satisfying quality of service to passengers, companies operating suburban trains seek to implement timetables that perform well despite the occurrence of random disturbances. However, due to some specificities of Mass Transit, the standard approaches for robust train timetabling do not apply in that context. In this paper, we present a data-based stochastic simulation method for assessing the efficiency of train timetables for dense traffic areas. We describe models for the random variabilities that occur daily on such networks, and use them in a stochastic simulation algorithm. Results are presented on a saturated line of Paris suburban network.

IEEE ITSC 2020, 23rd IEEE International Conference on Intelligent Transportation Systems, Rhodes, GRECE, 20-/09/2020 - 23/09/2020; This work contributes to the operational design of dynamic HOV lanes on French urban freeways. We propose a framework to estimate the acceptable dynamic maximal speed of the HOV lane, knowing the current origin speed. The aim is to ensure safety of lane changes while maintaining an acceptable duration of the period before gap acceptance. It is recognized that the longer the time a driver waits before accepting a gap, the lower he/she places the threshold for next gap acceptance. That may lead to inappropriate behaviors, such as unsafe lane changes. The key variables for lane changing realization are speeds and headways on origin and target lanes (estimated here with individual loop data). Both variables play an important role in lane change maneuver safety. We present in this paper two lane change realization models combining simple rules and observed data, one pessimistic and one optimistic. From this, we determine the maximal and minimal boundaries of two indicators: the number of rejected gaps and the waiting time to safe lane change. Those indicators lead to an appraisal of lane change realization safety. The results tend to show that a speed regulation lower than usual speed may increase safety while not degrading the service provided by this lane to its specific users. Having in mind the perspective of lanes dedicated to connected automated cruise control vehicles, we extend the proposed method to provide an analysis of the impact of uniforms inter-vehicular time gap on lane change safety.

KES AMSTA 2020, 14th International Conference on Agents and Multi-Agent Systems: Technology and Applications - Virtual conference, Split, CROATIE, 17-/06/2020 - 19/06/2020; Today policy makers face an ever more complex traffic system. While they need to ensure smooth traffic flows in the cities, at the same time an acceptable level of service must be provided in remote areas. Autonomous taxis (AT) offer the opportunity to manage car traffic at low operational cost and they can be appropriate alternatives for the driven vehicles. In this paper, we propose a multi-agent system to find the best dispatching strategy for a fleet of AT. In the dispatching process, we aim to satisfy both passengers and providers objectives and priorities. To be able to apply the method on large-scale networks, we introduce a clustering method to cluster the requests every minute and then solve the assignment problem for the requests inside each cluster. As the network congestion can have significant impacts on the vehicles speed and travel time especially considering the private cars that are driving in the system besides the taxis, an agent-based simulation platform is used to simulate the function of the AT fleet. We use the trip-based Macroscopic Fundamental Diagram (MFD) to simulate the time evolution of traffic flows on the road network and update the traffic situation in the system every second to represent the real traffic dynamics. We address the problem for a large city scale of 80 km2 (Lyon city in France) with more than 480,000 trips over 4 hours period containing the morning peak. The experimental results with real data show that the proposed multi-agent system is efficient in terms of serving all the requests in a short time satisfying both passengers and providers objectives.

Over the past century, cities around the world have experienced rapid growth in their populations. As a consequence, the capacity of their transport networks has been exceeded, causing traffic congestion problems that are associated to economic losses, high levels of air pollution and the emission of greenhouse gases. It is in this context that it is important to implement traffic regulation strategies to mitigate congestion. Traffic simulators are essential tools for developing and testing such strategies. The traffic conditions in the transport network are consequence of the travellers' choices at any given time, but at the same time, the traffic conditions on the network influence the decisions of the travellers. This relationship between travellers' decisions and network traffic is the main idea in traffic assignment, which traffic simulators use to produce network traffic patterns. It is in this sense that understanding and approximating the behaviour of travellers is indispensable to forecast the states of a transportation network, which could help design regulations to alleviate congestion. The objective of this thesis is to find route choice models that scale-up at network level, i.e., models that predict the choices of travellers over the diversity of situations found in a transport network. The approach in this thesis to investigate travellers' behaviour in transportation networks is through computer-based experiments at large scale, for which a platform named the Mobility Decision Game (MDG), has been developed. The MDG permits to observe the choices of the participants on a diverse set of scenarios (OD pairs and routes) with varying traffic conditions and travel time information. In this thesis, the experiments focus on the route choices of uni modal car trips that are based on the map of the city of Lyon, France. To attain the objective of this thesis, firstly a methodology to find OD pairs that are representative of the whole network is proposed. The representative OD experiments to obtain data to estimate choice models that generalize to the various OD pair configurations in the network. Secondly, the choices of participants in the experiments are analyzed from the rational and boundedly rational behaviour perspectives in order to establish the principle that best describe their choices. Finally, the choice models are assessed in terms of their predictive accuracy. This thesis is part of a European ERC project entitled MAGnUM: Multiscale and Multimodal Traffic Modeling Approach for Sustainable Management of Urban Mobility., Au cours du siècle dernier, les villes du monde entier ont connu une croissance rapide de leur population. En conséquence, la capacité de leurs réseaux de transport a été dépassée, entrainant des problèmes de congestion du trafic qui sont à leur tour associés à des pertes économiques, à des niveaux élèves de pollution atmosphérique et à l'émission de gaz à effet de serre. C'est dans ce contexte qu'il est important de mettre en oeuvre des stratégies de régulation du trafic pour atténuer les encombrements. Les simulateurs de trafic sont essentiels au développement et au test de telles stratégies. Les conditions de circulation dans le réseau de transport sont la conséquence des choix des voyageurs `a un moment donné, mais en même temps, les conditions de circulation sur le réseau influencent les décisions des voyageurs. Cette relation entre les décisions des voyageurs et le trafic réseau est l'idée principale de l'attribution du trafic, que les simulateurs de trafic utilisent pour produire des modèles de trafic réseau. C'est en ce sens que la compréhension et l'approximation du comportement des voyageurs sont indispensables pour prévoir les états d'un réseau de transport, ce qui pourrait aider à concevoir des réglementations pour réduire la congestion. Lorsqu'ils voyagent dans une ville, les voyageurs sont confrontés à plusieurs décisions concernant (i) l'activité à entreprendre, (ii) la destination du voyage, (iii) le mode de transport, (iv) l'heure de départ et (v) l'itinéraire. Ces décisions individuelles façonnent les états du trafic dans le réseau de transport `a un moment donné. Cependant, en même temps que les états du réseau sont la conséquence des choix des voyageurs, les caractéristiques et les états de trafic du réseau influencent les décisions des voyageurs. On pourrait penser, par exemple, aux voyageurs qui adaptent leur heure de départ et leurs choix d'itinéraire en anticipant des embouteillages sur la route habituelle. Cette relation bidirectionnelle entre le comportement des voyageurs et la dynamique du trafic est l'idée centrale des problèmes d'affectation du trafic et elle est étudiée sous deux angles différents. Le premier est lié au problème d'affectation du trafic, dans lequel les modèles de trafic au niveau du réseau sont obtenus à la suite des interactions entre les choix de tous les voyageurs dans le réseau de transport, sur la base d'hypothèses générales sur l'équilibre du réseau. Le second est lié aux choix des voyageurs, avec un accent particulier sur l'identification et la classification des facteurs qui influencent leurs choix. La motivation de cette thèse est d'obtenir des modèles de choix à la fois cohérents avec le comportement observé des voyageurs et généralisant bien au niveau du réseau à grande échelle. En d'autres termes, des modèles qui font des prédictions précises dans toutes les paires origine-destination (OD) dans un réseau de transport urbain, mais qui sont estimés avec des observations sur un nombre limité de paires OD. L'idée générale est de mettre en oeuvre, à l'avenir, ces modèles de choix dans des algorithmes d'affectation dynamique du trafic, et ainsi d'obtenir des schémas de trafic simulés dans une perspective comportementale. Les modèles de choix sont déduits des données collectées par des expériences informatiques réalisées sur une plateforme informatique Mobility Decision Game (MDG). Cependant, comme les choix des voyageurs ne peuvent pas être observés dans un si grand nombre de paires de OD et d'itinéraires que l'on trouve dans un réseau urbain, une m?méthodologie pour collecter les observations des choix des voyageurs par des expériences informatiques est d'abord nécessaire. Les modèles estimés avec des données provenant de cette méthodologie devraient pouvoir évoluer, d'un petit ensemble de paires de OD observables, jusqu'au niveau du réseau complet

The objective of this thesis is to find route choice models that scale-up at network level, i.e., models that predict the choices of travellers over the diversity of situations found in a transport network. The approach in this thesis to investigate travellers' behaviour in transportation networks is through computer-based experiments at large scale, for which a platform named the Mobility Decision Game (MDG), has been developed. The MDG permits to observe the choices of the participants on a diverse set of scenarios (OD pairs and routes) with varying traffic conditions and travel time information. In this thesis, the experiments focus on the route choices of uni-modal car trips that are based on the map of the city of Lyon, France. To attain the objective of this thesis, firstly a methodology to find OD pairs that are representative of the network is proposed. The representative OD pairs are used in route choice experiments to obtain choice models that generalise to the various OD configurations in the network. Secondly, the choices of participants in the experiments are analysed from the rational and boundedly rational behaviour perspectives, in order to establish the principle that best describe their choices. Finally, the choice models are assessed in terms of their predictive accuracy. This thesis is part of a European ERC project entitled MAGnUM: Multiscale and Multimodal Traffic Modeling Approach for Sustainable Management of Urban Mobility.; L'objectif de cette thèse est de trouver des modèles de choix d'itinéraire qui évoluent au niveau du réseau, c'est-à-dire des modèles qui rapprochent les choix des voyageurs sur la diversité des situations rencontrées dans un réseau de transport. L'approche de cette thèse pour étudier le comportement des voyageurs dans les réseaux de transport passe par des expériences informatiques à grande échelle, pour lesquelles une plateforme nommée Mobility Decision Game (MDG) a été développée. Le MDG permet d'observer les choix des participants sur un ensemble diversifié de scénarios (paires OD et itinéraires) avec des conditions de circulation et des informations de temps de trajet variables. Dans cette thèse, les expériences se concentrent sur les choix d'itinéraire des trajets en voiture qui sont basés sur la carte de la ville de Lyon, France. Pour atteindre l'objectif de cette thèse, une méthodologie de recherche de couples OD représentatifs du réseau est tout d'abord proposée. Les paires OD représentatives sont utilisées dans les expériences de choix de route pour obtenir des modèles de choix qui se généralisent aux différentes configurations OD dans le réseau. Deuxièmement, les choix des participants aux expériences sont analysés du point de vue du comportement rationnel et borné, afin d'établir le principe qui décrit le mieux leurs choix. Enfin, les modèles de choix sont évalués en fonction de leur précision prédictive. Cette thèse fait partie d'un projet européen ERC intitulé MAGnUM: Approche de modélisation du trafic multi-échelle et multimodale pour la gestion durable de la mobilité urbaine.

Over the past century, cities around the world have experienced rapid growth in their populations. As a consequence, the capacity of their transport networks has been exceeded, causing traffic congestion problems that are associated to economic losses, high levels of air pollution and the emission of greenhouse gases. It is in this context that it is important to implement traffic regulation strategies to mitigate congestion. Traffic simulators are essential tools for developing and testing such strategies. The traffic conditions in the transport network are consequence of the travellers' choices at any given time, but at the same time, the traffic conditions on the network influence the decisions of the travellers. This relationship between travellers' decisions and network traffic is the main idea in traffic assignment, which traffic simulators use to produce network traffic patterns. It is in this sense that understanding and approximating the behaviour of travellers is indispensable to forecast the states of a transportation network, which could help design regulations to alleviate congestion. The objective of this thesis is to find route choice models that scale-up at network level, i.e., models that predict the choices of travellers over the diversity of situations found in a transport network. The approach in this thesis to investigate travellers' behaviour in transportation networks is through computer-based experiments at large scale, for which a platform named the Mobility Decision Game (MDG), has been developed. The MDG permits to observe the choices of the participants on a diverse set of scenarios (OD pairs and routes) with varying traffic conditions and travel time information. In this thesis, the experiments focus on the route choices of uni modal car trips that are based on the map of the city of Lyon, France. To attain the objective of this thesis, firstly a methodology to find OD pairs that are representative of the whole network is proposed. The representative OD experiments to obtain data to estimate choice models that generalize to the various OD pair configurations in the network. Secondly, the choices of participants in the experiments are analyzed from the rational and boundedly rational behaviour perspectives in order to establish the principle that best describe their choices. Finally, the choice models are assessed in terms of their predictive accuracy. This thesis is part of a European ERC project entitled MAGnUM: Multiscale and Multimodal Traffic Modeling Approach for Sustainable Management of Urban Mobility.; Au cours du siècle dernier, les villes du monde entier ont connu une croissance rapide de leur population. En conséquence, la capacité de leurs réseaux de transport a été dépassée, entrainant des problèmes de congestion du trafic qui sont à leur tour associés à des pertes économiques, à des niveaux élèves de pollution atmosphérique et à l'émission de gaz à effet de serre. C'est dans ce contexte qu'il est important de mettre en oeuvre des stratégies de régulation du trafic pour atténuer les encombrements. Les simulateurs de trafic sont essentiels au développement et au test de telles stratégies. Les conditions de circulation dans le réseau de transport sont la conséquence des choix des voyageurs `a un moment donné, mais en même temps, les conditions de circulation sur le réseau influencent les décisions des voyageurs. Cette relation entre les décisions des voyageurs et le trafic réseau est l'idée principale de l'attribution du trafic, que les simulateurs de trafic utilisent pour produire des modèles de trafic réseau. C'est en ce sens que la compréhension et l'approximation du comportement des voyageurs sont indispensables pour prévoir les états d'un réseau de transport, ce qui pourrait aider à concevoir des réglementations pour réduire la congestion. Lorsqu'ils voyagent dans une ville, les voyageurs sont confrontés à plusieurs décisions concernant (i) l'activité à entreprendre, (ii) la destination du voyage, (iii) le mode de transport, (iv) l'heure de départ et (v) l'itinéraire. Ces décisions individuelles façonnent les états du trafic dans le réseau de transport `a un moment donné. Cependant, en même temps que les états du réseau sont la conséquence des choix des voyageurs, les caractéristiques et les états de trafic du réseau influencent les décisions des voyageurs. On pourrait penser, par exemple, aux voyageurs qui adaptent leur heure de départ et leurs choix d'itinéraire en anticipant des embouteillages sur la route habituelle. Cette relation bidirectionnelle entre le comportement des voyageurs et la dynamique du trafic est l'idée centrale des problèmes d'affectation du trafic et elle est étudiée sous deux angles différents. Le premier est lié au problème d'affectation du trafic, dans lequel les modèles de trafic au niveau du réseau sont obtenus à la suite des interactions entre les choix de tous les voyageurs dans le réseau de transport, sur la base d'hypothèses générales sur l'équilibre du réseau. Le second est lié aux choix des voyageurs, avec un accent particulier sur l'identification et la classification des facteurs qui influencent leurs choix. La motivation de cette thèse est d'obtenir des modèles de choix à la fois cohérents avec le comportement observé des voyageurs et généralisant bien au niveau du réseau à grande échelle. En d'autres termes, des modèles qui font des prédictions précises dans toutes les paires origine-destination (OD) dans un réseau de transport urbain, mais qui sont estimés avec des observations sur un nombre limité de paires OD. L'idée générale est de mettre en oeuvre, à l'avenir, ces modèles de choix dans des algorithmes d'affectation dynamique du trafic, et ainsi d'obtenir des schémas de trafic simulés dans une perspective comportementale. Les modèles de choix sont déduits des données collectées par des expériences informatiques réalisées sur une plateforme informatique Mobility Decision Game (MDG). Cependant, comme les choix des voyageurs ne peuvent pas être observés dans un si grand nombre de paires de OD et d'itinéraires que l'on trouve dans un réseau urbain, une m?méthodologie pour collecter les observations des choix des voyageurs par des expériences informatiques est d'abord nécessaire. Les modèles estimés avec des données provenant de cette méthodologie devraient pouvoir évoluer, d'un petit ensemble de paires de OD observables, jusqu'au niveau du réseau complet

Coupling a traffic microsimulation with an emission model is a means of assessing fuel consumptions and pollutant emissions at the urban scale. Dealing with congested states requires the efficient capture of traffic dynamics and their conditioning for the emission model. Two emission models are investigated here: COPERT IV and PHEM v11. Emission calculations were performed at road segments over 6 min periods for an area of Paris covering 3 km2. The resulting network fuel consumption (FC) and nitrogen oxide (NOx) emissions are then compared. This article investigates: (i) the sensitivity of COPERT to the mean speed definition, and (ii) how COPERT emission functions can be adapted to cope with vehicle dynamics related to congestion. In addition, emissions are evaluated using detailed traffic output (vehicle trajectories) paired with the instantaneous emission model, PHEM. COPERT emissions are very sensitive to mean speed definition. Using a degraded speed definition leads to an underestimation ranging from -13% to -25% for fuel consumption during congested periods (from -17% to -36% respectively for NOx emissions). Including speed distribution with COPERT leads to higher emissions, especially under congested conditions (+13% for FC and +16% for NOx). Finally, both these implementations are compared to the instantaneous modeling chain results. Performance indicators are introduced to quantify the sensitivity of the coupling to traffic dynamics. Using speed distributions, performance indicators are more or less doubled compared to traditional implementation, but remain lower than when relying on trajectories paired with the PHEM emission model.; Le couplage d'une microsimulation de trafic avec un modèle d'émissions est un moyen d'évaluer les consommations de carburant et les émissions polluantes à l'échelle urbaine. Le traitement des états congestionnés nécessite l'intégration efficace de la dynamique du trafic et son conditionnement pour le modèle d'émissions. Deux modèles d'émissions sont étudiés ici: COPERT IV et PHEM v11. Des calculs d'émissions ont été réalisés sur des tronçons routiers sur des périodes de 6 minutes pour une zone de Paris de 3 km2. La consommation de carburant (FC) et les émissions d'oxydes d'azote (NOx) du réseau qui en résultent sont ensuite comparées. Cet article analyse : (i) la sensibilité de COPERT à la définition de la vitesse moyenne, et (ii) la manière dont les fonctions d'émission de COPERT peuvent être adaptées pour faire face à la dynamique des véhicules liée à la congestion. De plus, les émissions sont évaluées à l'aide de données détaillées sur le trafic (trajectoires des véhicules) jumelées au modèle des émissions instantanées, PHEM. Les émissions de COPERT sont très sensibles à la définition de la vitesse moyenne. L'utilisation d'une définition de vitesse dégradée conduit à une sous-estimation allant de -13% à -25% pour la consommation de carburant en période de congestion (de -17% à -36% respectivement pour les émissions NOx). L'inclusion de la distribution de vitesse avec COPERT entraîne une augmentation des émissions, surtout en conditions de congestion (+13% pour le FC et +16% pour le NOx). Enfin, ces deux implémentations sont comparées aux résultats de la chaîne de modélisation instantanée. Des indicateurs de performance sont introduits pour quantifier la sensibilité du couplage à la dynamique du trafic. En utilisant des distributions de vitesse, les indicateurs de performance sont plus ou moins doublés par rapport à la mise en oeuvre traditionnelle, mais restent inférieurs à ceux qui reposent sur des trajectoires couplées au modèle d'émission PHEM.